X-Git-Url: https://vcs.fsf.org/?p=exim.git;a=blobdiff_plain;f=src%2Fsrc%2Fexigrep.src;h=bb994d7697f35ead2e493a354677eba7aeb24c37;hp=e824bb41ca1e7e51d7d985dc8ea1c13050e9ee5a;hb=44bc8f0c2f3576b46bd6df1b818cb29eaf84df5b;hpb=55ee9ee384af22d5e7aa995868d0d90c4d7026ad diff --git a/src/src/exigrep.src b/src/src/exigrep.src index e824bb41c..bb994d769 100644 --- a/src/src/exigrep.src +++ b/src/src/exigrep.src @@ -1,9 +1,8 @@ #! PERL_COMMAND -w -# $Cambridge: exim/src/src/exigrep.src,v 1.2 2004/12/21 11:28:38 ph10 Exp $ use strict; -# Copyright (c) 2004 University of Cambridge. +# Copyright (c) 2007-2015 University of Cambridge. # See the file NOTICE for conditions of use and distribution. # Except when they appear in comments, the following placeholders in this @@ -28,6 +27,10 @@ use strict; # appears to be compressed, it is passed through zcat. We can't just do this # for all files, because zcat chokes on non-compressed files. +# Performance optimized in 02/02/2007 by Jori Hamalainen +# Typical run time acceleration: 4 times + + use Getopt::Std qw(getopts); use POSIX qw(mktime); @@ -38,7 +41,7 @@ use POSIX qw(mktime); sub seconds { my($year,$month,$day,$hour,$min,$sec,$tzs,$tzh,$tzm) = - $_[0] =~ /^(\d{4})-(\d\d)-(\d\d)\s(\d\d):(\d\d):(\d\d)(?>\s([+-])(\d\d)(\d\d))?/; + $_[0] =~ /^(\d{4})-(\d\d)-(\d\d)\s(\d\d):(\d\d):(\d\d)(?>\s([+-])(\d\d)(\d\d))?/o; my $seconds = mktime $sec, $min, $hour, $day, $month - 1, $year - 1900; @@ -55,17 +58,27 @@ return $seconds; # This subroutine processes a single line (in $_) from a log file. Program # defensively against short lines finding their way into the log. -my (%saved, %id_list, $pattern, $queue_time); +my (%saved, %id_list, $pattern, $queue_time, $insensitive, $invert); + +# If using "related" option, have to track extra message IDs +my $related; +my $related_re=''; +my @Mids = (); sub do_line { + +# Convert syslog lines to mainlog format, as in eximstats. + +if (!/^\d{4}-/o) { $_ =~ s/^.*? exim\b.*?: //o; } + return unless - my($date,$entry) = /^(\d{4}-\d\d-\d\d \d\d:\d\d:\d\d (?:[+-]\d{4} )?)(.*)/; + my($date,$id) = /^(\d{4}-\d\d-\d\d \d\d:\d\d:\d\d (?:[+-]\d{4} )?)(?:\[\d+\] )?(\w{6}\-\w{6}\-\w{2})?/o; # Handle the case when the log line belongs to a specific message. We save # lines for specific messages until the message is complete. Then either print -# discard. +# or discard. -if (my($id) = $entry =~ /^(\w{6}\-\w{6}\-\w{2})/) +if (defined $id) { $saved{$id} = '' unless defined($saved{$id}); @@ -73,29 +86,45 @@ if (my($id) = $entry =~ /^(\w{6}\-\w{6}\-\w{2})/) $saved{$id} .= $_; - # Are we interested in this id ? + # Are we interested in this id ? Short circuit if we already were interested. - $id_list{$id} = 1 if /$pattern/io; + if ($invert) + { + $id_list{$id} = 1 if (!defined($id_list{$id})); + $id_list{$id} = 0 if (($insensitive && /$pattern/io) || /$pattern/o); + } + else + { + if (defined $id_list{$id} || + ($insensitive && /$pattern/io) || /$pattern/o) + { + $id_list{$id} = 1; + get_related_ids($id) if $related; + } + elsif ($related && $related_re) + { + grep_for_related($_, $id); + } + } # See if this is a completion for some message. If it is interesting, # print it, but in any event, throw away what was saved. - if ($entry =~ - /(?:Completed|rejected (?:by local_scan|by non-SMTP ACL|after DATA))/) + if (index($_, 'Completed') != -1 || + index($_, 'SMTP data timeout') != -1 || + (index($_, 'rejected') != -1 && + /^(\d{4}-\d\d-\d\d \d\d:\d\d:\d\d (?:[+-]\d{4} )?)(?:\[\d+\] )?\w{6}\-\w{6}\-\w{2} rejected/o)) { - if ($saved{$id} =~ /^(\d{4}-\d\d-\d\d \d\d:\d\d:\d\d ([+-]\d{4} )?)(\w{6}\-\w{6}\-\w{2})/) + if ($queue_time != -1 && + $saved{$id} =~ /^(\d{4}-\d\d-\d\d \d\d:\d\d:\d\d ([+-]\d{4} )?)/o) { my $old_sec = &seconds($1); my $sec = &seconds($date); - delete $id_list{$id} if $id_list{$id} && $sec - $old_sec <= $queue_time; - } - - if ($id_list{$id}) - { - delete $id_list{$id}; - print "$saved{$id}\n"; + $id_list{$id} = 0 if $id_list{$id} && $sec - $old_sec <= $queue_time; } + print "$saved{$id}\n" if ($id_list{$id}); + delete $id_list{$id}; delete $saved{$id}; } } @@ -103,18 +132,84 @@ if (my($id) = $entry =~ /^(\w{6}\-\w{6}\-\w{2})/) # Handle the case where the log line does not belong to a specific message. # Print it if it is interesting. -elsif ($entry =~ /$pattern/io) { print "$_\n"; } +elsif ( ($invert && (($insensitive && !/$pattern/io) || !/$pattern/o)) || + (!$invert && (($insensitive && /$pattern/io) || /$pattern/o)) ) + { print "$_\n"; } } +# Rotated log files are frequently compressed and there are a variety of +# formats it could be compressed with. Rather than use just one that is +# detected and hardcoded at Exim compile time, detect and use what the +# logfile is compressed with on the fly. +# +# List of known compression extensions and their associated commands: +my $compressors = { + gz => { cmd => 'zcat', args => '' }, + bz2 => { cmd => 'bzcat', args => '' }, + xz => { cmd => 'xzcat', args => '' }, + lzma => { cmd => 'lzma', args => '-dc' } +}; +my $csearch = 0; + +sub detect_compressor_bin + { + my $ext = shift(); + my $c = $compressors->{$ext}->{cmd}; + $compressors->{$ext}->{bin} = `which $c 2>/dev/null`; + chomp($compressors->{$ext}->{bin}); + } + +sub detect_compressor_capable + { + my $filename = shift(); + map { &detect_compressor_bin($_) } keys %$compressors + if (!$csearch); + $csearch = 1; + return undef + unless (grep {$filename =~ /\.(?:$_)$/} keys %$compressors); + # Loop through them, figure out which one it detected, + # and build the commandline. + my $cmdline = undef; + foreach my $ext (keys %$compressors) + { + if ($filename =~ /\.(?:$ext)$/) + { + # Just die if compressor not found; if this occurrs in the middle of + # two valid files with a lot of matches, error could easily be missed. + die("Didn't find $ext decompressor for $filename\n") + if ($compressors->{$ext}->{bin} eq ''); + $cmdline = $compressors->{$ext}->{bin} ." ". + $compressors->{$ext}->{args}; + last; + } + } + return $cmdline; + } + +sub grep_for_related { + my ($line,$id) = @_; + $id_list{$id} = 1 if $line =~ m/$related_re/; +} + +sub get_related_ids { + my ($id) = @_; + push @Mids, $id unless grep /\b$id\b/, @Mids; + my $re = join '|', @Mids; + $related_re = qr/$re/; +} # The main program. Extract the pattern and make sure any relevant characters # are quoted if the -l flag is given. The -t flag gives a time-on-queue value -# which is an additional condition. +# which is an additional condition. The -M flag will also display "related" +# loglines (msgid from matched lines is searched in following lines). -getopts('lt:',\my %args); -$queue_time = $args{'t'}? $args{'t'} : -1; +getopts('Ilvt:M',\my %args); +$queue_time = $args{'t'}? $args{'t'} : -1; +$insensitive = $args{'I'}? 0 : 1; +$invert = $args{'v'}? 1 : 0; +$related = $args{'M'}? 1 : 0; -die "usage: exigrep [-l] [-t ] []...\n" +die "usage: exigrep [-I] [-l] [-M] [-t ] [-v] []...\n" if ($#ARGV < 0); $pattern = shift @ARGV; @@ -129,11 +224,16 @@ if (@ARGV) foreach (@ARGV) { my $filename = $_; - if ($filename =~ /\.(?:COMPRESS_SUFFIX)$/) + if (-x 'ZCAT_COMMAND' && $filename =~ /\.(?:COMPRESS_SUFFIX)$/o) { open(LOG, "ZCAT_COMMAND $filename |") || die "Unable to zcat $filename: $!\n"; } + elsif (my $cmdline = &detect_compressor_capable($filename)) + { + open(LOG, "$cmdline $filename |") || + die "Unable to decompress $filename: $!\n"; + } else { open(LOG, "<$filename") || die "Unable to open $filename: $!\n"; @@ -147,8 +247,11 @@ if (@ARGV) else { do_line() while (); } -# At the end of processing all the input, print any uncompleted data +# At the end of processing all the input, print any uncompleted messages. -for (keys %id_list) { print "+++ $_ not completed +++\n$saved{$_}\n;" } +for (keys %id_list) + { + print "+++ $_ has not completed +++\n$saved{$_}\n"; + } # End of exigrep